公开(公告)号：US09116135B2

公开(公告)日：2015-08-25

申请号：US13469193

申请日：2012-05-11

Applicant: Wen-Wei Cheng ,  Ming-Kai Hsueh ,  Wen-Chi Lo

Inventor： Wen-Wei Cheng ,  Ming-Kai Hsueh ,  Wen-Chi Lo

IPC: G06K9/48 ,  G01N21/956 ,  G01N21/95 ,  G06T7/00 ,  H01L27/146

CPC classification number: G01N21/956 ,  G01N21/9501 ,  G06T7/0004 ,  G06T2207/30148 ,  H01L27/14601 ,  H01L27/14603

Abstract: A surface pattern detecting method includes: capturing a surface image of a sample element to be detected, wherein the surface image containing N grayscale pixels and wherein the N is a positive integer; selecting f×N pixels with small grayscale value based on a selection ratio f in order to define a pattern zone and further calculating a pattern mean of the pattern zone based on pixel amount and grayscale value of the pattern zone; selecting f×N pixels with big grayscale value in order to define a background zone and further calculating a background mean of the background zone based on pixel amount and grayscale value of the background zone; and determining whether the surface image has a defect based on the pattern means of the pattern zone and the background mean of the background zone.

Abstract translation: 表面图案检测方法包括：捕获要检测的样本元素的表面图像，其中所述表面图像包含N个灰度像素，并且其中N是正整数; 基于选择比f选择具有小灰度值的f×N个像素，以便基于图案区域的像素量和灰度值来定义图案区域并进一步计算图案区域的图案均匀度; 选择具有大灰度值的f×N个像素，以便基于背景区域的像素量和灰度值定义背景区域并进一步计算背景区域的背景平均值; 以及基于图案区域的图案装置和背景区域的背景平均值来确定表面图像是否具有缺陷。